Uma modelagem baseada no PSO para estimativas das taxas de emissão de particulados em um acidente nuclear

Abstract

Nuclear power plants are complex facilities and require special safety care, especially in the case of an accident with radiation release into the atmosphere. Therefore, different approaches have been investigated and proposed to define and reduce uncertainties, in the modeling of terms source and radioactive plumes, which are released into the atmosphere. The determination of the source term and its release rate in a nuclear emergency are useful for decision-making. This research aims to estimate the rate of particulate contaminants emitted by multiple sources, whose values are unknown, using the values identified by the receptors distributed around the sources. To this end, a model of Particle Swarm Optimization (PSO) is applied together with a mathematical model of Gaussian dispersion, and the program PSO Dispersion Model (PSODM) is created. In PSODM, each particle is a solution for the emission rate of the sources. The output is the solution that best represents the emission rates of multiple sources, that is, it is the solution of better fitness, obtained by the smallest error between the measured values in the real receivers and the measured values in the calculated receivers. The PSODM, presented efficacy in the solution of the problem approached showing its feasibility of application in problems of this nature in the nuclear area.